If AI Is Going to Refuse Me Anyway, Why Not Use a Chinese AI?
The other day, while writing a blog post, an AI flatly refused to help me.
No matter how many times I rephrased it, the answer was always the same: “I can’t write that.”
Well, I get it. AI companies care a great deal about ethics. It’s a product used all over the world, so if it wrote something strange and caused a public backlash, they would be the ones in trouble. I understood that.
But after being turned down, a thought occurred to me.
If it can’t write about sensitive topics anyway, is there really any reason to stick with an American AI?
For Everyday Work, a Chinese AI Is Good Enough
Chinese AI has gotten quite good lately. Its performance has largely caught up with the latest American models, and it is much cheaper.
Drafting emails, summarizing documents, making small fixes to code. If that’s all you’re asking for, a Chinese AI works just fine.
Of course, there are things a Chinese AI can’t write either. Ask about the Tiananmen Square incident and it will change the subject. Ask it to poke fun at President Xi Jinping, and it will probably refuse with everything it has.
In other words, both American AI and Chinese AI have things they won’t write. The only difference is what counts as taboo.
American AI is strict about ethics. Chinese AI is strict about the Communist Party. If what I want to write is not criticism of the Communist Party but silly stories about everyday life, a Chinese AI might actually be easier to get along with.
Your Data Gets Taken Either Way
Of course, Chinese AI comes with the risk that “the information you enter might be taken by the Chinese Communist Party.”
China has a law requiring companies to cooperate with the state’s intelligence work. Data is also stored on servers inside China. Some countries have even banned the use of Chinese AI in government agencies.
But once you start down that road, America isn’t much different.
America also has laws that let the government make tech companies hand over data when needed. Taken by the Communist Party, or taken by the White House. The only difference is whether it goes to Beijing or Washington.
And right now, Washington has a president who loves publicly attacking critics by name. Which one is the lesser evil is a pretty close contest.
Tolerant of Criticism Aimed at Itself
So maybe a Chinese AI is an option. That’s what I was thinking.
By the way, I had an American AI help me with the draft of this article.
The very American AI I was complaining about, calling it hard to use because it’s so strict about ethics, wrote that criticism smoothly without refusing once.
It seems to be strict about human ethics, but quite tolerant when the criticism is about itself. I kind of like that about it.